Hi folks, I have been trying to get my head around dualities, and hit a stumbling block right away. Duality hypotheses are framed by thinking about perturbative expansions of interaction lagrangians ( at least when thinking about s-dualities). In such expansions, the first term is regarded as the 'classical' term, and what that means in the path integral approach is that the first term corresponds to the classical trajectory permitted by the interaction. But then its clear the the perturbative expansion in mind in these discussions of duality cannot be the expansion about the interaction coupling and that Feynman diagrams illustrate, because there the first term is the interaction-free term - not the one describing the classically-permitted trajectory of the interacting system. So can someone tell me what I'm missing when people like Sen discuss dualities in terms of these expansions ( see eg here http://www.iisc.ernet.in/currsci/dec251999/articles20.htm) ?! I appreciate this is going to be really obvious, but any help would be much appreciated!